Appendix 3. Risk of Bias Summary of Included Studies
Study: Kumar 2007 Methods
Randomized controlled trial
Participants 24 Eyes. Visual acuity ≤ 6/60, diffuse macular edema, HbA1C > 7.5%, study based in India.
Excluded patients with fundus fluorescein angiogram (FFA) evidence of macular ischemia, vitreomacular traction, cataract surgery within 1 year, previous vitrectomy, previous panretinal photocoagulation (PRP) in 12months, previous grid laser in 6 months, uncontrolled diabetes, hypertension or chronic renal failure
Interventions
Pars plana vitrectomy (PPV) with dye-assisted internal limiting membrane (ILM) peel or grid LASER. No triamcinolone used
Outcomes
Early Treatment of Diabetic Retinopathy Study (ETDRS) visual acuity (VA), macular thickness, macular volume at 6 months
Bias
Authors judgemen t
Support for judgement Random sequence generation
(selection bias) Unclear
Randomisation method not mentioned Allocation concealment (selection
bias) Unclear Randomisation concealed by sealed
envelop Blinding of participants and
personnel (performance bias) High risk
Patients not masked (vitrectomy or laser) Blinding of outcome assessment
(detection bias) Unclear
No mention if assessors were masked Incomplete outcome data (attrition
bias) Low risk
25 patients assessed, 24 eyes analysed Selective reporting (reporting bias) Low risk
All outcomes reported
Other bias Low risk
Study: Patel 2006 Methods
Randomized controlled trial Participants
20 patients recruited, 7 vitrectomy patients, 8 laser patients
Clinically significant macular edema (CSME) depsite previous macular laser (less than 2years), Snellen acuity 6/15- 6/60 (65- 35 letters)
Excluded macular ischemia Interventions
Vitrectomy (ILM not peeled) (7 patients) or macular laser (8 patients) Outcomes
ETDRS VA, optical coherence tomography (OCT), and FFA at baseline and 12 months
Bias Authors
judgement Support for judgement Random sequence generation
(selection bias)
Low risk
Randomisation performed by separate department in same hospital.
Comment: Probably done Allocation concealment (selection
bias)
High risk
Concealment method not described Blinding of participants and
personnel (performance bias)
High risk
Patients can not be masked to intervention
Blinding of outcome assessment (detection bias)
Low risk
Masked observers Incomplete outcome data (attrition
bias)
Low risk
20 patients recruited, 15 analysed- similar attrition in each group Selective reporting (reporting bias) Low risk
All outcomes reported
Other bias Low risk
Study: Thomas 2005 Methods
Randomized controlled trial Participants
Diabetic macular edema with no macular traction Previous macular laser
VA 6/12 or worse
Excluded ischemic maculopathy Interventions
19 patients randomised to vitrectomy with ILM peel compared to 21 control patients undergoing further macular laser
Outcomes
Best corrected logarithm of the minimum angle of resolution (LogMAR) VA and mean OCT central macular thickness
Bias Authors
judgement Support for judgement Random sequence generation
(selection bias)
Low risk
Randomization prepared by independent medical statistician
Allocation concealment (selection bias)
Low risk
Sealed, numbered opaque envelopes used: probably adequate
Blinding of participants and personnel (performance bias)
High risk
Unable to mask patients Blinding of outcome assessment
(detection bias)
Unclear
risk No mention if assessors were masked Incomplete outcome data (attrition
bias)
Low risk
Similar loss to follow up in both groups (vitrectomy group loss 4/19, laser control group loss 3/18)
Selective reporting (reporting bias) Low risk
All outcomes reported
Study: Yanyali 2005 Methods
Randomised controlled trial Participants
12 patients, 24 eyes
Bilateral diabetic macular edema Type 2 diabetes
No vitreomacular traction No previous laser
Interventions
12 eyes vitrectomy with ILM peeling Fellow 12 eyes undergo macular laser Outcomes
Best corrected LogMAR VA and mean OCT central macular thickness
Bias Authors
judgement Support for judgement Random sequence generation
(selection bias)
High risk
Method of randomisation not explained Allocation concealment (selection
bias)
High risk
Method of concealment not explained Blinding of participants and
personnel (performance bias)
High risk
Participant not masked Blinding of outcome assessment
(detection bias)
Unclear
risk No mention if assessors were masked Incomplete outcome data (attrition
bias)
Low risk
Full follow up Selective reporting (reporting bias) Low risk
All outcomes reported
Other bias Low risk
Study: Yanyali 2006 Methods
Randomized controlled trial Participants
10 patients, 20 eyes
Bilateral diabetic macula edema Type 2 diabetes
All had previous grid laser and now considered 'unresponsive to further laser'
No vitreomacular traction Interventions
10 eyes underwent vitrectomy with ILM peeling 10 fellow eyes followed observation alone Outcomes
OCT retinal thickness and visual acuity
Bias Authors
judgement Support for judgement Random sequence generation
(selection bias)
High risk
Method of randomization not explained Allocation concealment (selection
bias)
High risk
Method of concealment not explained Blinding of participants and
personnel (performance bias)
High risk
Unable to mask patients Blinding of outcome assessment
(detection bias)
Low risk
Assessor was masked to treatment groups
Incomplete outcome data (attrition bias)
Low risk
No patients lost to follow up Selective reporting (reporting bias) Low risk
All outcomes reported
Other bias Low risk
